[Decision] Sequencer Deployment Options

[RossiNYC]

Objective

Verify options for the Shannon sequencer deployment and present to PNF/Community for decision.

Origin Document

TL;DR: PNF/DAO needs to decide who/how the Shannon Sequencer will be deployed AND the process required for making the decision (e.g. Community Vote)
See: https://www.notion.so/poktdao/Sequencer-Options-Shannon-bc11a37f685343188dd12b1a86fe459b for more context.

SKIP Cost Estimation: https://skip-protocol.notion.site/Skip-Ecosystem-Development-Kit-EDK-Pricing-for-POKT-6f9598a3ff0c411bbf8d54830a520ef8 (Proposal withdrawn - no longer an option)

Caldera RFP Response: https://docs.google.com/document/d/1P3WOGzVB92kR6VgS3GsBId-uzHI2URJM7AC6BGRgV7U/edit

Goals

Verify options (e.g. Does Rollkit/Celestia support CosmoSDK x/Gov)
If outsourcing to a 3rd party - Identify 3rd Parties under consideration (e.g. SKIP)

Deliverables

• Clear documentation on the process of making the Sequencer decision
• Un-bias explainer document for all vetted options including decision matrix/factors (e.g. Cost, Time to Deploy, etc.)
• Risks/Mitigations for each option
• Recommendation Document for DAO voting

Non-goals / Non-deliverables

The actual decision, this Issue lays out what is needed to get to the decision, not the decision itself.
The implementation details of the final decision

Creator: @RossiNYC
Co-Owners: Ben @ PNF

[Olshansk]

@RossiNYC Could you also add an epic/iteration on when we need to look into this.

[RossiNYC]

PNF has reviewed the SKIP cost documentation

B3n has been added to the TG chat and we're trying to organize a call for this week. Meanwhile, reviewing other options for comparison/de-risking.

[RossiNYC]

Additionally, I looked at shared sequencer options that are part of the Celestia ecosystem:

• Astria - looks like they have a testnet ready to go - they are the furthest ahead from what I can gather from websites
• Radius - docs are there but not much else, they recently raised, looks like a work in progress
• Fairblock also just raised 2.5M in Oct, I like the idea a lot but they are probably 12-18 months out from a testnet.

[Olshansk]

Posting a quick update:

1. The team at skip.money withdrew their proposal (see screenshot below).
2. The team at caldera is already experimenting with rollkit for other chains (see screenshot below).

My recommendations for next steps are:

1. Leverage the support from Skip to have a meeting and collect requirements of what we need.
2. Meet with the team from Caldera to understand the overlap/differences of their offerings and the requirements we collect based on (1).
3. Use the results of (1) & (2) to formalize our requirements, get an estimate of costs and then the following:

• Put together a proposal (via the DAO) to let the community (i.e. Pocket validator runners) bid on the project
• Enable other RaaS providers bid on the project alongside our community.

@RossiNYC Wanted to know what you think and if you'd be able to lead the charge on this?

---

[Olshansk]

@RossiNYC Please post our sequencer requirements here once we have them ironed out. Thank you!

[Olshansk]

Posting @jacklaing's notes after meeting with the Espresso Systems team below.

---

• Decentralized sequencer uses hotstuff, reaches finality after 3 blocks, and scales to any block time - won’t be a bottleneck.
• They have a DA layer called tiramisu that runs in parallel to consensus, but teams can use other DA like Celestia.
• They’ve been giving Celestia feedback on Rollkit. Sequencers in Celestia are DA-aware because they’re required to pay fees to DA providers but Espresso’s sequencer intentionally holds minimal state so doesn’t do fee tracking for what the sequencer needs to pay third-parties (only minimal fee tracking of what is needed to cover Espresso’s own costs).
• Decentralized Sequencer Roadmap: smaller testnet December (2 node operators), larger testnet January, mainnet by end of Q1.
• Will be doing a code freeze followed by inter-team code audit (they have 3 teams - sequencer, consensus, and cryptography) and an external audit.
  Partners: Injective, Caldera, Altlayer, Lumos, Off-chain Labs
• Haven’t landed on pricing yet - having pricing discovery calls with partners
• Support: shared Slack channels with partners or Telegram groups, whichever the partner prefers.
• Docs are kind of barebones at the moment but will improve in the next few weeks as we get closer to testnet.
• Other partner opportunities: seemed interested in Pocket being used by their partners to send transactions to the sequencer (via RPC), subject to partner interest

[Olshansk]

Sharing the RFP @RossiNYC put together here.

[Olshansk]

Summarizing some notes from a conversation I had with some of our advisors on this matter:

• Astria
  • Less mature but very promising
  • Likely to provide a lot more hands-on support at a good price
• Caldera
  • Heavily EVM focused but experimenting with Rollkit
  • Will provide some hands-on support
• Conduit
  • Heavily focused on the EVM
  • Might not be able to provide a lot of hands-on support
• Strangelove
  • Cosmos native team that is very skilled and very experienced but likely to be very expensive
• Binary
  • Cosmos native team that would be open to helping us

[RossiNYC]

Caldera RFP Response: https://docs.google.com/document/d/1P3WOGzVB92kR6VgS3GsBId-uzHI2URJM7AC6BGRgV7U/edit

I also updated the notion page https://www.notion.so/poktdao/Sequencer-Options-Shannon-bc11a37f685343188dd12b1a86fe459b with their answers and included the requirements table we sent out with the RFP so we can line up the responses vs. req.

[Olshansk]

Closing this out as the decision is to:

1. Go with Astria
2. Use Caldera or run our own if timelines misalign.